Earth Syst. Sci. Data, 10, 1281–1300, 2018https://doi.org/10.5194/essd-10-1281-2018© Author(s) 2018. This work is distributed underthe Creative Commons Attribution 4.0 License.

Mediterranean Sea Hydrographic Atlas: towardsoptimal data analysis by including time-dependent

statistical parametersAthanasia Iona1,2, Athanasios Theodorou2, Sylvain Watelet3, Charles Troupin3, Jean-Marie Beckers3,

and Simona Simoncelli41Hellenic Centre for Marine Research, Institute of Oceanography, Hellenic National Oceanographic Data

Centre, 46,7 km Athens Sounio, Mavro Lithari P.O. Box 712 19013 Anavissos, Attica, Greece2University of Thessaly, Department of Ichthyology & Aquatic Environment,

Laboratory of Oceanography, Fytoko Street, 38 445, Nea Ionia Magnesia, Greece3University of Liège, GeoHydrodynamics and Environment Research, Quartier Agora,

Allée du 6-Août, 17, Sart Tilman, 4000 Liège 1, Belgium4Istituto Nazionale di Geofisica e Vulcanologia (INGV), Sezione di Bologna,

Via Franceschini 31, 40128 Bologna, ItalyCorrespondence: Athanasia Iona (sissy@hnodc.hcmr.gr)

Received: 30 January 2018 – Discussion started: 2 February 2018Revised: 13 April 2018 – Accepted: 16 April 2018 – Published: 16 July 2018

Abstract. The goal of the present work is to provide the scientific community with a high-resolution atlas oftemperature and salinity for the Mediterranean Sea based on the most recent datasets available and contributeto the studies of the long-term variability in the region. Data from the pan-European marine data infrastructureSeaDataNet were used, the most complete and, to our best knowledge, best quality dataset for the MediterraneanSea as of today. The dataset is based on in situ measurements acquired between 1900 and 2015. The atlas consistsof horizontal gridded fields produced by the Data-Interpolating Variational Analysis, in which unevenly spatialdistributed measurements were interpolated onto a 1/8◦× 1/8◦ regular grid on 31 depth levels. Seven differ-ent types of climatological fields were prepared with different temporal integration of observations. Monthly,seasonal and annual climatological fields have been calculated for all the available years, seasonal to annual cli-matologies for overlapping decades and specific periods. The seasonal and decadal time frames have been chosenin accordance with the regional variability and in coherence with atmospheric indices. The decadal and specific-period analysis was not extended to monthly resolution due to the lack of data, especially for the salinity. TheData-Interpolating Variational Analysis software has been used in the Mediterranean region for the SeaDataNetand its predecessor Medar/Medatlas Climatologies. In the present study, a more advanced optimization of theanalysis parameters was performed in order to produce more detailed results. The past and present states of theMediterranean region have been extensively studied and documented in a series of publications. The purposeof this atlas is to contribute to these climatological studies and get a better understanding of the variability ontimescales from months to decades and longer. Our gridded fields provide a valuable complementary source ofknowledge in regions where measurements are scarce, especially in critical areas of interest such as the MarineStrategy Framework Directive (MSFD) regions and subregions. The dataset used for the preparation of the atlasis available from https://doi.org/10.12770/8c3bd19b-9687-429c-a232-48b10478581c.

The climatologies in netCDF are available at the following sources: annual climatol-ogy (https://doi.org/10.5281/zenodo.1146976), seasonal climatology for 57 running decades(https://doi.org/10.5281/zenodo.1146938), seasonal climatology (https://doi.org/10.5281/zenodo.1146953),annual climatology for 57 running decades (https://doi.org/10.5281/zenodo.1146957), seasonal cli-matology for six periods (https://doi.org/10.5281/zenodo.1146966), annual climatology for six periods(https://doi.org/10.5281/zenodo.1146970), monthly climatology (https://doi.org/10.5281/zenodo.1146974).

Published by Copernicus Publications.

1282 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

1 Introduction

In oceanography, a climatology is defined as a set of grid-ded fields that describe the mean state of the ocean proper-ties over a given time period. It is constructed by the analy-sis of in situ historical data sets and has many applicationssuch as initialization of numerical models and quality con-trol of observational data in real time and delayed mode, andit is used as a baseline for comparison to understand howthe ocean is changing. The Mediterranean Sea is among themost interesting regions in the world because it influencesthe global thermohaline circulation and plays an importantrole in regulating the global climate (Lozier et al., 1995;Béthoux et al., 1998; Rahmstorf, 1998). It is therefore es-sential to improve our understanding of its dynamics and itsprocess variability. It is a semi-enclosed sea divided by theSicily Strait in two geographical basins, the western Mediter-ranean and the eastern Mediterranean, and is characterizedby peculiar topographic deep depressions where nutrient-richdeep-water masses are stored for long time (Manca et al.,2004). It is a concentration basin, where evaporation ex-ceeds precipitation. The general mean circulation pattern isdriven by the continuous evaporation, heat flux exchangeswith the atmosphere, the wind stress and the water mass ex-changes between its basins and subbasins, as described inRobinson et al. (2001). In the surface layer (0–150 m) thereis an inflow of warm and relatively fresh Atlantic water (AW;S≈ 36.5 ppt, T ≈ 15 ◦C) which is modified along its pathto the eastern basin following a general cyclonic circula-tion. The intermediate layers (150–600 m) are dominated bythe saline Levantine Intermediate Water (LIW; S ≈ 38.4 ppt,T ≈ 13.5 ◦C), regularly formed at the Levantine basin. LIWis one of the most important Mediterranean water masses be-cause it constitutes the higher percentage of the outflow fromthe Gibraltar Strait towards the Atlantic Ocean. In the deeplayers, there are two main thermohaline cells. Deep watersformed via convective events in the northern regions of thewestern Mediterranean (WMDW – Western MediterraneanDeep Water; S ≈ 38.44–38.46 ppt, T ≈ 12.75–13.80 ◦C) inthe Gulf of Lion and in the northern regions of the easternMediterranean (EMDW – Eastern Mediterranean Deep Wa-ters) in the Adriatic (S ≈ 38.65 ppt, T ≈ 13.0 ◦C) and Cre-tan seas (S ≈ 39 ppt, T ≈ 14.8 ◦C). On top of this large-scale general pattern are superimposed several subbasin-scale and mesoscale cyclonic and anticyclonic motions dueto topographic constraints and internal processes. The firstclimatological studies of the Mediterranean go back to 1966with Ovchinnikov who carried out a geostrophic analysisto compute the surface circulation. The circulation featureswere describing a linear, stationary ocean. In the 1980s and1990s, through a comprehensive series of observational stud-ies and experiments in the western Mediterranean (La Vi-olette, 1990; Millot, 1987, 1991, 1999), the subbasin and

mesoscale patterns were discovered and the crucial role ofeddies in modifying the mean climatological circulation andmixing properties inside the different subbasins which in-clude the Tyrrhenian, Ligurian, and Alboran seas was empha-sized (Bergamasco and Malanotte-Rizzoli, 2010). In 1987,Guibout constructed charts with the typical structures ob-served at sea, but with limited climatological characteristicas this atlas was based on the quasi-synoptic information ofthe selected cruise used. In 1982, an international researchgroup was formed, called POEM (Physical Oceanography ofthe Eastern Mediterranean, 1984), under the auspices of theIOC/UNESCO and of CIESM (Commission Internationalepour l’Exploration Scientifique de la Méditerranée), which isfocused on the description of the phenomenology of the east-ern Mediterranean, by analyzing historical data and collect-ing new data (Malanotte-Rizzoli and Hecht, 1988). However,the focus was on the eastern Mediterranean Sea, as there wasvery little knowledge of this basin compared to the westernMediterranean and other world regions, and still there wasno global data set to describe the eastern Mediterranean andits interaction with the western part efficiently. It is worthnoting here that it was the POEM cruises that provided theobservational evidence of the profound changes in the east-ern Mediterranean deep circulation between the late 1980sand mid-1990s. During this phenomenon, known as East-ern Mediterranean Transient (EMT) the deep water produc-tion in the eastern Mediterranean shifted from the Adriatic tothe Aegean. The bottom layers of the eastern Mediterraneanwere replaced by waters warmer but also much saltier andhence denser than the previous Eastern Mediterranean DeepWaters (Malanotte-Rizzoli et al., 1999; Lascaratos et al.,1999; Klein et al., 1999; Roether et al., 1996). The EMT af-fected not only the thermohaline characteristics of intermedi-ate and deep water masses of eastern Mediterranean but alsothe western Mediterranean deep water production (Schröderet al., 2006; Schroeder et al., 2016). The EMT “signature”can be captured to the climatological references of a regionand can be used to improve the near-real-time control proce-dures of operational and delayed mode data (Manzella et al.,2013). Hecht et al. (1988), by analyzing hydrographic mea-surements in the southeastern Levantine Basin, succeeded indescribing the climatological water masses of the region andidentifying their seasonal variations. In the western Mediter-ranean, Picco (1990) conducted an important climatologi-cal study and constructed a climatological atlas analyzingaround 15 000 hydrological profiles from various sources forthe period 1909–1987. In 1982, Levitus published the firstclimatological atlas of the world ocean at a 1◦ resolution,using temperature, salinity, and oxygen data from CTD, bot-tle stations, mechanical, and expendable bathythermographsof the previous 80 years. An objective analysis at standardlevels was carried out to compute the climatology. Since1994, a new version is released every 4 years. The current

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/

A. Iona et al.: Mediterranean Sea Hydrographic Atlas 1283

version that includes the Mediterranean (World Ocean At-las 2013, V2; Locarnini et al., 2013) is a long-term set ofobjectively analyzed climatologies of temperature, salinity,oxygen, phosphate, silicate, and nitrate for annual, seasonal,and monthly periods for the world ocean. It also includes as-sociated statistical fields of observed oceanographic profiledata from which the climatologies were computed. In ad-dition to increased vertical resolution, the 2013 version hasboth 1 and 1/4◦ horizontal resolution versions available forannual and seasonal temperature and salinity for 6 decades,as well as monthly for the decadal average. Brasseur et al.(1996) introduced a new method to reconstruct the three-dimensional fields of the properties of the MediterraneanSea. Seasonal and monthly fields were analyzed using a vari-ational inverse method (VIM) to generate the climatolog-ical maps, instead of the objective analysis introduced byGandin (1966) and Bretherton et al. (1976) on the meteo-rology and oceanography which was widely used by then.More than 34 000 CTD and bottle data were used and in-tegrated in the so-called MED2 historical database. Com-parison of the results obtained with both methods showedthat VIM was mathematically equivalent but numericallymore efficient than the objective analysis. The method hasbeen adopted by the Medar/Medatlas and its successor Sea-DataNet Project.

At the beginning of the 2000s, many research projectsand monitoring activities have produced large amounts ofmultidisciplinary in situ hydrographic and biochemical datafor the whole Mediterranean Sea, but still the data werefragmented and inaccessible from the scientific community.The aim of the EU/MAST Medar/Medatlas Project, 2001(http://www.ifremer.fr/medar/, last access: 9 July 2018) wasto rescue and archive the dispersed data through a wide co-operation of countries and produce an atlas for 12 core pa-rameters: temperature and salinity, dissolved oxygen, hy-drogen sulfur, alkalinity, phosphate, ammonium, nitrite, ni-trate, silicate, chlorophyll and pH. Gridded fields have beencomputed on a 1/4◦ grid resolution using the VIM andthe DIVA tool developed by the GHER group of the Uni-versity of Liège. Interannual and decadal variabilities oftemperature and salinity were computed as well. The at-las is available at http://modb.oce.ulg.ac.be/backup/medar/contribution.html (last access: 9 July 2018) and on CD-Rom(MEDAR Group, 2002). The atlas contains a selection of fig-ures and three-dimensional fields in netCDF. The EU FP7SeaDataNet project, 2006–2011 and 2012–2016 (the suc-cessor of the EU/MAST Medar/Medatlas), has integratedhistorical, multidisciplinary data on a unique, standardizedonline data management infrastructure and provides value-added aggregated datasets and regional climatologies basedon these aggregated datasets for all the European sea basins.SeaDataNet has adopted the VIM method and the DIVA soft-ware tool. Temperature and salinity monthly climatologieshave been produced on a 1/8◦ grid resolution (Simoncelliet al., 2015a, 2016, https://doi.org/10.12770/90ae7a06-8b08-

4afe-83dd-ca92bc99f5c0). These climatologies are basedon the V1.1 historical data collection of publicly avail-able temperature and salinity in situ profiles (Simoncelliet al., 2014, https://doi.org/10.12770/90ae7a06-8b08-4afe-83dd-ca92bc99f5c0) spanning the time period 1900–2013(Simoncelli et al., 2015a).

1.1 Objectives

The objective of this study is the computation of an improvedatlas compared with the existing products in the Mediter-ranean Sea using the latest developments of the DIVA toolwith the aim to contribute to the better representation of theclimatological patterns and understanding of the long-termvariability of the regional features of the basin. The orig-inality of this product compared to the SeaDataNet clima-tology (both products use similar techniques) is the highertemporal resolution, up to decadal, and more advanced cal-ibration of the analysis parameters for improving the re-sults and the representation of general circulation patternsat timescales smaller than the climatic means. Besides theWOA13 decadal periods climatologies (1955–1964, 1965–1974, 1975–1984, 1985–1994, 1995–2004, and 2005–2012)the Medar/Medatlas provides interannual and decadal grid-ded fields and therefore it is compared with the present at-las. The present product has two major improvements: higherspatial and vertical resolutions and error fields that accom-pany all the analysis results, therefore allowing a more reli-able assessment of the results. The advantage of this prod-uct in relation to the WOA13 climatology is the higher spa-tial resolution (up to 1/8◦× 1/8◦) longitude–latitude gridsused and the higher temporal resolution as the analysis usesrunning decades instead of successive decades. Another im-portant difference with respect to all previous existing cli-matologies in the Mediterranean is that additional and morerecent data are used. The atlas constructed consists of cli-matological fields (called climatology hereafter) to depictthe “mean” state of the Mediterranean region at monthly,seasonal (winter: January–March; spring: April–June; sum-mer: July–September; fall: October–December), and annualscale. Additionally, climatological fields at seasonal and an-nual scale for 57 running decades have been produced to de-pict the interannual and decadal variability of the system andreveal the decadal trends.

One additional period from 2000–2015 was produced forthose users or applications who are interested to reference tothe latest data and new platform types such as Argo floats.This period is provided with the five previous successivedecadal periods, e.g., 1950–1959, 1960–1969, 1970–1979,1980–1989, 1990–1999, and 2000–2015. In summary, thefollowing gridded products are available:

– monthly climatological gridded fields obtained by an-alyzing all monthly data of the whole period 1950 to2015;

www.earth-syst-sci-data.net/10/1281/2018/ Earth Syst. Sci. Data, 10, 1281–1300, 2018

1284 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

– seasonal climatological gridded fields obtained by ana-lyzing all monthly data of the whole period from 1950to 2015 falling within each season;

– seasonal gridded fields obtained by analyzing all datafalling within each season for each of 57 seasons overthe running decades from 1950–1959 to 2006–2015;

– seasonal gridded fields obtained by analyzing datafalling within each season for six periods: 1950–1959, 1960–1969, 1970–1979, 1980–1989, 1990–1999,2000–2015;

– annual climatology obtained by analyzing all data (re-gardless month or season) for the whole period from1950 to 2015;

– annual gridded fields obtained by analyzing all data re-gardless month or season for each of the 57 seasons overthe running decades from 1950–1959 to 2006–2015;

– annual gridded fields obtained by analyzing data fallingwith each season for six periods, 1950–1959, 1960–1969, 1970–1979, 1980–1989–1990–1999, 2000–2015.The first five periods coincide with the correspondingdecades.

The climatologies provided cannot rely on sufficient high-frequency and high-resolution data to allow mesoscale fea-tures, which play an important role and modify the large-scale flow fields, to be resolved (Robinson et al., 2001). Wefocus thus on the seasonal and decadal variations. This timefiltering also results in a spatial filter as later shown by thespatial correlations found in the data. The spatial scales thedata can capture are of the order of 300–350 km at the sur-face, much larger than the Rossby radius of deformationscale (10–15 km) associated with mesoscale motions (30–80 km, Robinson et al., 2001). These mesoscale features arethus filtered out from the analysis and hence the numericalgrids we will use only need to resolve the large scales. Thesame holds for the output files, where there is no reason tosave at very high resolution (much smaller than the deforma-tion radius) as in any case the analysis provides large-scalefields.

The atlas covers the geographical region 30–46◦ N,6.25◦W–36.5◦ E on 31 standard depth levels from 0to 4000 m. The fields are stored in netCDF files. Theatlas is accessible in netCDF from the Zenodo plat-form using the following DOIs: annual climatology(https://doi.org/10.5281/zenodo.1146976), seasonal clima-tology for 57 running decades from 1950–1959 to 2006–2015 (https://doi.org/10.5281/zenodo.1146938); seasonalclimatology (https://doi.org/10.5281/zenodo.1146953), an-nual climatology for 57 running decades from 1950–1959 to2006–2015 (https://doi.org/10.5281/zenodo.1146957);seasonal climatology for six periods: 1950–1959,1960–1969, 1970–1979, 1980–1989, 1990–1999,

2000–2015 (https://doi.org/10.5281/zenodo.1146966);annual climatology for six periods: 1950–1959, 1960–1969, 1970–1979, 1980–1989, 1990–1999, 2000–2015(https://doi.org/10.5281/zenodo.1146970); monthly clima-tology (https://doi.org/10.5281/zenodo.1146974).

Tables 1 and 2 give the state of the art of the existing cli-matologies in the Mediterranean Sea.

2 Data

2.1 Data source

The SeaDataNet historical temperature and salinity datacollection V2 for the Mediterranean Sea was used(Simoncelli et al., 2015b, http://sextant.ifremer.fr/record/8c3bd19b-9687-429c-a232-48b10478581c/, last access: 9July 2018). The collection includes 213 542 temperature and138 691 salinity profiles from in situ measurements cover-ing the 1911–2015 period, which corresponds to all open-access data available through the European SeaDataNet Ma-rine Data Infrastructure (www.seadatanet.org, last access: 9July 2018). These datasets were collected from 102 dataproviders, are quality controlled and archived in 32 marineand oceanographic data centers, and distributed into the in-frastructure by 27 SeaDataNet partners. The data range from9.25◦W to 37◦ E and include a part of the Atlantic (not in-cluded in the atlas) and the Marmara Sea. Figure 1 shows thelocations of the profiles and Table 3 the main instrumentsof the collection. Users have to register at the Marine-ID(https://users.marine-id.org, last access: 9 July 2018) to getan account for downloading the SeaDataNet V2 data collec-tion. Registration is done only once and thereafter users canhave access not only to SeaDataNet but also to all EMODnetand Copernicus Marine Data services.

The profiles in the SeaDataNet collection come fromin situ observations collected with various instruments andplatforms such as CTD and bottles data by discrete watersamplers operated by research vessels or other smaller ves-sels, bathythermographs (mechanical - MBT - or expendable- XBT). Table 3 summarizes the major numbers for temper-ature and salinity profiles and their temporal spanning as de-rived from the station type, the instrument, and the platformtype.

There are 21 682 vertical profiles of temperature and salin-ity from CTD and bottle stations with no information on theinstrument used. These figures correspond to profiles beforethe quality control and the processing of data for preparationof climatology.

The bathythermograph data were maintained in the col-lection despite the incorrect fall rate and the resulting warmbias in the measurements (Wijffels et al., 2008). Yet, there isnot a XBT/MBT correction for the Mediterranean as for theglobal ocean (https://www.nodc.noaa.gov/OC5/XBT_BIAS/xbt_bias.html, last access: 9 July 2018). These data signifi-cantly improve the geographical coverage of the temperature

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/

A. Iona et al.: Mediterranean Sea Hydrographic Atlas 1285

Figure 1. Geographical distribution of (a) temperature and (b) salinity profiles of the SeaDataNet V2 historical data collection (Simoncelliet al., 2015b, http://sextant.ifremer.fr/record/8c3bd19b-9687-429c-a232-48b10478581c/, last access: 9 July 2018).

records during the 1950–1985 period and reduce the result-ing analysis error due to data gaps. Therefore, it was decidedto keep them despite their known (small) bias.

2.1.1 Preparation for the analysis

The following operations were applied to the initial data be-fore performing the analysis:

1. Data were extracted at the selected depth levels and tem-poral frames (months, seasons, decades, and periods).

2. Spatial binning was performed to improve the qualityof the parameters (correlation length and signal-to-noiseratio) optimization by averaging a part of the highly cor-related observations (taken by a research vessel in a re-

stricted area, for instance). Spatial binning was appliedto the data only during the parameter optimization stepbut then the original data (and nonbinned) were used forthe analysis itself.

3. Data outside the analysis domain were excluded.

4. Weights were applied to the data to reduce the influenceof a large number of data located in a small area withina short period. This is particularly useful for time seriesdata. The characteristic length of weighting was set tobe equal to 0.08◦ and the characteristic time of weight-ing was set to be equal to 90 days for the seasonal anal-ysis and 30 days for the monthly analysis. The scaleswere chosen according to the spatial and temporal reso-lution of the analysis.

www.earth-syst-sci-data.net/10/1281/2018/ Earth Syst. Sci. Data, 10, 1281–1300, 2018

1286 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

Table 1. Overview of characteristics of existing T/S climatologies in Mediterranean Sea. OSD=Ocean Station Data,CTD= conductivity–temperature–depth, MBT=mechanical–digital–micro bathythermograph, XBT= expendable bathythermograph,SUR= surface, APB= autonomous pinniped bathythermograph, MRB=moored buoy data, PFL= profiling float data, DRB= drifting buoy,UOR= undulating oceanographic recorder, GLD= glider.

Climatology World Ocean Atlas2013

Medar/Medatlas SeaDataNet Present

Date 2013 2002 2015 2017Instruments/platforms OSD, High-resolution

CTD, MBT, XBT,SUR, APB, MRB,PFL, DRB, UOR, GLD

MBT, XBT, discretewater samplers, CTD,thermistor chains

MBT, XBT, discretewater samplers, CTD,thermistor chains, ther-mosalinographs, DRB,PFL, MRB

MBT, XBT, discretewater samplers, CTD,thermistor chains, ther-mosalinographs, DRB,PFL, MRB

Horizontal extent Global Mediterranean, BlackSea

Mediterranean Sea Mediterranean Sea

Parameters Temperature, salinity,density beta version,conductivity, dis-solved oxygen, percentoxygen saturation,apparent oxygenutilization, silicate,phosphate, nitrate

Temperature, salin-ity, alkalinity, pH, dis-solved oxygen,ammonium, nitrite,nitrate, phosphate, sili-cates, chlorophyll-a,hydrogen sulfide

Temperature, salinity Temperature, salinity

Horizontal resolution 1/4◦× 1/4◦ 1/4◦× 1/4◦ 1/8◦× 1/8◦ 1/8◦× 1/8◦

Vertical extent 0–1500 m, 0–5500 m 0–4000 m 0–5500 m 0–5500 mVertical levels 57 levels for monthly

fields, 102 levels for an-nual, seasonal fields

25 33 33

Temporal data coverage 1864–2013 1890–2000 1900–2013 1900–2015Temporal resolution Climatic, monthly,

seasonal, averagedperiods, decadal(1955–1964, 1965–1974, 1975–1984,1985–1994, 1995–2004, 2005–2012years)

Climatic, monthly,seasonal, interannual∗,running 3 years∗, run-ning 5 years, runningdecades∗

Monthly Climatic, monthly, sea-sonal, running decades,averaged periods

No error fields are available for the Medar/Medatlas, interannual∗, running 3-years∗ and running decades∗ computations.

Table 2. Analysis parameters for variational inverse method (VIM) implementation in the Mediterranean.

Climatology Medar/Medatlas SeaDataNet PresentMethod VIM DIVA (VIM) DIVA (VIM)

Correlation length Constant Constant VariableSignal-to-noise ratio Variable Constant VariableBackground field Seminormed Seminormed SeminormedDetrending No No Yes (for background fields)Observation weighting No No Yes

2.1.2 Vertical interpolation

The observations do not have a uniform vertical distribu-tion and a vertical interpolation of the profiles into stan-dard depths is needed prior to the gridding. The DIVA toolhas embedded the weighted parabolic interpolation method

(Reiniger and Ross, 1968) into its workflow (Troupin et al.,2010). This method is widely used in oceanographic clima-tologies, such as World Ocean Atlas 2013 (WOA13) (Lo-carnini et al., 2013; Zweng et al., 2013) and Medar/Medatlas(MEDAR Group, 2002), as it creates less vertical instabili-ties. Only “good” data were used, e.g., data with quality con-

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/

A. Iona et al.: Mediterranean Sea Hydrographic Atlas 1287

Figure 2. Number of observations per month for (a) temperatureand (b) salinity.

trol flag 1 and 2 according to the SeaDataNet QC flag scale(SeaDataNet Group, 2010). The interpolation was performedonto 31 International Oceanographic Data and InformationExchange (IODE) standards depths: [0, 5, 10, 20, 30, 50, 75,100, 125, 150, 200, 250, 300, 400, 500, 600, 700, 800, 900,1000, 1100, 1200, 1300, 1400, 1500, 1750,2000, 2500, 3000,3500, 4000]. These are the same depths as Simoncelli et al.(2014) used in the existing SeaDataNet V1.1 climatology,thus allowing quick visual intercomparisons. The WOA13climatological fields offer a higher vertical resolution (87depth levels from 0 to 4000 m), thus facilitating higher res-olution models or more accurate quality control for obser-vational data. However, the monthly WOA13 fields in theMediterranean extend only down to 1500 m. In the next re-leases of the present atlas, the vertical resolution will be in-

creased. This version rather focuses on the improvement ofthe horizontal and temporal resolutions.

2.1.3 Quality control

Currently the SeaDataNet V2 collection is the most com-plete and quality-controlled data set for the MediterraneanSea, released within SeaDataNet project. Prior to its release,it has undergone extended quality control according to theSeaDataNet standards and strategy such as elimination ofduplicates, broad range checks for detecting outliers, spikes,density inversion, zero values. Regional experts performedthe quality checks in close cooperation with the data origina-tors and the responsible data centres ensuring the best resultas described in Simoncelli et al. (2015b). The quality con-trol is done with the use of the Ocean Data View tool (ODV,Schlitzer, 2002, https://odv.awi.de, last access: 9 July 2018).However, in order to avoid the influence of extremes (but notnecessarily erroneous values) in the climatology, the follow-ing additional checks were applied to the data. The Mediter-ranean Sea was divided in 27 rectangular areas defined byManca et al. (2004), which exhibit the typical subregionalfeatures. In each area, a mean profile was calculated at thestandard depths. Values lying outside±3 standard deviationsaround the mean were considered as “outliers” and excludedfrom the DIVA analysis. The percentage of data excludedfrom the analysis was 0.8 % for the temperature and 0.9 % forthe salinity values. This filter with the standard deviation wasapplied twice at the data sets that were used as backgroundfields and for the optimization of the analysis parameters (thecorrelation length, signal-to-noise ratio and variance of thebackground field). In the second run of the standard devia-tion filter, the amount of excluded data was 0.5 and 0.7 %respectively.

2.1.4 Data temporal distributions

There is an expected seasonal distribution with more temper-ature and salinity profiles during summer and autumn as wellas at the surface layers compared to the deeper ones (Fig. 2).However, the monthly and yearly distributions reveal a greatbias for September and October and for the years 2008 and2009. This is due to the temporary presence of 51 063 ther-mosalinograph data, at 2 and 3 m depth, essentially locatednear the Strait of Gibraltar and in the Alboran Sea. Such ir-regular distribution in time has to be taken into account whencomputing the climatological fields otherwise the spatial in-terpolations will be biased towards the values of these highdata coverage during these 2 months. Therefore, a detrend-ing method was applied to the data prior to the DIVA analy-sis to remove this effect of uneven distribution in time. Thedetrending tool is provided by the DIVA software. The de-trending was applied to all background fields (see paragraphbelow) and the monthly, seasonal, and annual climatologies.It was not applied to running decades and the 6 decadal pe-

www.earth-syst-sci-data.net/10/1281/2018/ Earth Syst. Sci. Data, 10, 1281–1300, 2018

1288 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

Figure 3. Averaged profiles of correlation length (a, b) and signal-to noise ratio (c, d) for temperature and salinity.

riod climatologies in order not to remove any of the long-term trends in temperature and salinity.

2.1.5 Data weighting

The influence of the uneven distribution in space where alarge number of data points are concentrated in a very smallarea and within a very short period is controlled by apply-ing different weights to each of these data points. Indeed,such points cannot be considered independent in a clima-tological analysis. So rather than calculating a super obser-vation (similarly to binning), one can reduce the weight or,

in other words, increase the error attached to each individ-ual measurement. Points which are close in time and spacewill undergo such a treatment. The scales below which sucha weighting is done have a characteristic length of 0.08◦

(same unit as the data locations) and a characteristic timeof 1 month (30 days). Typical examples where the weightingis particularly profitable are the cases of the thermosalino-graph data (see previous paragraph) and time series data fromcoastal monitoring stations. Data weighting functionality isalso provided by the DIVA tool. Both data weighting and de-

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/

A. Iona et al.: Mediterranean Sea Hydrographic Atlas 1289

Table 3. Data distribution per instrument, station, and platform type.

Number of profilesTemperature Salinity Period

Instrument/station type Platform type

CTD Research vessel, moored surface buoy, fixed mooring, 45 914 42 420 1964–2013ships of opportunity, other type of vessels

Discrete water samplers Research vessel, other type of vessels 33 449 30 423 1912–2009Bathythermographs Research vessel, ships of opportunity, other type of vessels 44 336 1952–2014Thermosalinographs Research vessel, other type of vessels 51 063 29 747 2008–2009Salinity sensors, water Drifting subsurface float 8206 8207 2005–2015temperature sensorNot provided Research vessel, ships of opportunity, 21 682 21 682 1914–2013

other type of vessels, unknown

Figure 4. Example of observations distribution (a) and corresponding fields distributions depending on relative error threshold values, (b) anunmasked field, (c) a masked field with relative error threshold= 0.3, and (d) a masked field with relative error threshold= 0.5.

trending techniques have been applied for the first time in thecomputations for the Mediterranean Sea climatologies.

3 Method

3.1 The DIVA interpolation tool

The Data-Interpolating Variational Analysis (DIVA) is amethod designed to perform spatial interpolation (analysis)of sparse and heterogeneously distributed and noisy data intoa regular grid in an optimal way. The basic idea of the vari-

ational analysis is to determine a continuous field approxi-mating data and exhibiting small spatial variations. In otherwords, the target of the analysis is defined as the smoothestfields that respects the consistency with the observations anda priori knowledge of the background field over the domainof interest. To do so, a cost function that takes into accountthe distance between the reconstructed field and the observa-tions, and the regularity of the field is minimized. The solu-tion of the minimization problem is obtained through a finite-element technique (Rixen et al., 2000). The main advantageis that the computational cost is independent of the number

www.earth-syst-sci-data.net/10/1281/2018/ Earth Syst. Sci. Data, 10, 1281–1300, 2018

1290 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

Figure 5. Surface temperature climatology at 5 m in (a) January, (b) April, (c) July and (d) October.

of data points analyzed; instead it depends on the numberof degrees of freedom, i.e., on the size of the finite-elementmesh. The mesh takes into account the complexity of the ge-ometry of the domain without having to separate subbasinsprior to the interpolation, and the mesh automatically pro-hibiting correlations across land barriers. Among other ma-jor advantages of the method, DIVA can take into accountdynamic constraints allowing for anisotropic spatial correla-tion. Tools to generate the finite-element mesh are providedas well as tools to optimize the parameters of the analy-sis. The signal-to-noise ratio is optimized by a generalizedcross-validation (GCV) technique (Brankart and Brasseur,1996) while the correlation length is estimated by compar-ing the relation between the empirical data covariance andthe distance against its theoretical counterpart (Troupin et al.,2017). Along with the analysis gridded fields, DIVA alsoprovides error fields (Brankart and Brasseur, 1998; Rixenet al., 2000) based on the data coverage and their noise.The method computes gridded fields in two dimensions. Thethree-dimensional (x,y,z) and four-dimensional extensions(x,y,z, t) have been embedded into the interpolation schemewith an emphasis on creating climatologies. Detailed docu-mentation of the method can be found in Troupin et al. (2010,2012) and in the Diva User Guide (Troupin et al., 2017). Thecurrent atlas has been produced with the use of DIVA 4.6.11(Watelet et al., 2015a) with a Linux Ubuntu 16.04.3 operat-ing system.

3.2 Topography and coastlines

The domain where the interpolation has to be performed iscovered by a triangular finite-element mesh that follows thecoastline. The bathymetry used for coastline definitions andmesh generation is based on the General Bathymetric Chartof the Oceans (GEBCO) 1 min topography. Since the subse-quent analyses focus on much larger scales than 1 min, theresolution of the topography was downgraded to 5 min, stillfine enough to resolve topological features such as islandsand straits, yet coarse enough to be coherent with the scalesof interest. The step of the output grid was set to 1/8◦ for sim-ilar reasons. The geographical boundaries of the region wereset to 30–46◦ N, 6.25◦W–36.5◦ E. The depth contours thatdefine the two-dimensional horizontal planes where the in-terpolation takes place are the 31 standard depth levels from0 to 4000 m.

3.3 Finite-element mesh

Once the depth contour files at the specified 31 standarddepth levels were prepared, the mesh was generated usingan initial correlation length Lc equal to 0.5◦, which meansthe initial size Le of each finite triangular element is equalto 0.167◦ (Lc/3). The initial Lc scale was made as small asallowed by computing resources. This length scale is muchsmaller than the length scales for the analysis (typically afew degrees) and it ensures that the finite-element solution is

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/

A. Iona et al.: Mediterranean Sea Hydrographic Atlas 1291

Figure 6. Surface salinity climatology at 5 m for (a) winter, (b) spring, (c) summer, and (d) autumn.

Figure 7. Winter salinity at 100 m for four periods: (a) 1970–1979, (b) 1980–1989, (c) 1990–1999, (d) 2000–2015.

www.earth-syst-sci-data.net/10/1281/2018/ Earth Syst. Sci. Data, 10, 1281–1300, 2018

1292 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

solving the mathematical DIVA formulation with very highprecision (it does not mean that the analysis is the truth, butthat the numerical solution is actually close to the mathemat-ical solution of the variational formulation; in other words,the discretization does not add further errors into the analy-sis than the analysis method and the data themselves). Themesh was generated once and used in each repetition of theanalysis.

3.4 Interpolation parameters

3.4.1 Correlation length

The correlation length (Lc) is the radius of influence ofdata points. It can be determined objectively using a specificDIVA tool that takes into account the data distribution, orthe correlation length can be provided a priori by the useraccording to its (subjective) experience on expected patternsoccurring in the domain of interest. In the contrary to theSeaDataNet climatology where a constant value used in alldepths, in this atlas, the correlation length was defined asfollows. For the monthly climatology, the correlation lengthwas calculated for every month (for all years from 1950 to2015) and at each depth by fitting the empirical data co-variance as a function of distance by its theoretical function.It was then smoothed by applying a vertical filtering, simi-larly to Troupin et al. (2010). Then, these monthly correla-tion length profiles were averaged together in order to havea smooth transition from one month to another. Indeed, theestimated correlation lengths at the surface vary between 1.8(July) and 3.6◦ (December). For the seasonal and the annualclimatologies, the correlation length was calculated for everyseason and depth (from all years between 1950 and 2015), fil-tered vertically and then averaged into a single profile. Thisapproach yields slightly smaller correlation lengths than theaveraging per month and in combination with the rest of theanalysis parameters, the fields are not so smoothed and arecloser to the data. The mean profiles for temperature andsalinity are shown in Fig. 3a and c. There is a general in-crease from the surface to 2000 m except in a layer from 300to 600 m where it decreases. Also there is a decrease fromaround 2000 to 3000 m in both cases that can be attributedto the variability of the intermediate and deep waters in theMediterranean.

3.4.2 Signal-to-noise ratio

The signal-to-noise ratio (S / N) represents the ratio of thevariance of the signal to the variance of observational errors.When measuring a variable, there is always an uncertaintyon the value obtained. Noise does not only take into accountinstrumental errors (which are generally low), but it also in-cludes the following:

1. the representativeness errors, meaning that what onemeasures is not always what one intends to analyze(e.g., skin temperature, inadequate scales);

2. the synopticity errors, occurring when the measure-ments are assumed to be taken at the same time (e.g.,data from a cruise; Rixen et al., 2001).

Because of the multiple sources of error, a perfect fit to ob-servations is thus not advised. In the case of climatology,the main cause of error is not the instrumental error but therepresentativeness errors, which cannot easily be quantified.In this atlas, unlike the SeaDataNet (Simoncelli et al., 2014,2016) and other regional climatologies (Troupin et al., 2010),a variable S / N was defined following the same approach aswith the correlation length. The mean monthly and seasonalS / N profiles for temperature and salinity were vertically fil-tered. The remaining extremes were further filtered out man-ually by replacing them with the mean of the adjacent layers.The seasonal mean profile for temperature was further fil-tered out to a constant value from 1300 to 4000 m. In the first500 m (see Fig. 3b and d), both the monthly and seasonalprofiles display similar behaviors with a mean value rangingfrom 3.5 to 4.5, with an exception of the monthly mean valuefor salinity at 400 m.

3.5 Background field

The background field is the first guess of the gridded field(analysis) to reconstruct. First of all, this background is sub-tracted from data. Then, the variational analysis is performedby DIVA using these data anomalies as an input. Lastly, thebackground is added to the solution so that original data andfinal analyses are both expressed as absolute values. In areaswith very few or no data, the analysis tends to the backgroundfield since the data anomalies are close to zero (Brankartand Brasseur, 1998). In this atlas, a seminormed analysisfield (roughly speaking an analysis only capturing very largescales) was chosen as the background in order to guaranteethe solution is most realistic in areas void of data. Dependingon the type of climatology and gridded fields, two differentbackground fields were used: (a) a climatic seasonal field toaccount for seasonality from the original data, and (b) a cli-matic annual field to account for interannual variability. Thesame correlation lengths and signal-to-noise ratios were usedas for the atlas computations (see previous paragraph). Ana-lytically the following background fields were produced:

1. climatic seasonal seminormed analysis as backgroundfor the decadal seasonal fields of temperature and salin-ity;

2. climatic annual seminormed analysis as background forthe decadal annual fields of temperature and salinity;

3. climatic seasonal seminormed analysis as backgroundfor the periodical seasonal fields of temperature andsalinity;

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/

A. Iona et al.: Mediterranean Sea Hydrographic Atlas 1293

Figure 8. Winter salinity at 500 m for four periods: (a) 1970–1979, (b) 1980–1989, (c) 1990–1999, (d) 2000–2015.

Figure 9. Comparison of atlas temperature (a, c) and SeaDataNet climatology (b, d) between January at 10 m (a, b) and November at1000 m (c, d).

www.earth-syst-sci-data.net/10/1281/2018/ Earth Syst. Sci. Data, 10, 1281–1300, 2018

1294 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

Figure 10. Basin vertical averages of BIAS and RMSE between theatlas and SeaDataNet monthly temperature (a), and monthly salin-ity (b) climatologies.

4. climatic annual seminormed analysis as background forthe periodical annual fields of temperature and salinity;

5. climatic seasonal seminormed analysis as backgroundfor the monthly and seasonal climatology;

6. data mean value as background for the annual clima-tologies.

Additional monthly and seasonal climatologies were com-puted using the data mean value as background (the datamean value is subtracted from the data values).

The background climatic seasonal and annual fields weredetrended in order to remove any seasonal and interannualtrends that are due to nonuniform spatial data distribution(Capet et al., 2014). A characteristic example is the nonuni-form distribution of thermosalinographs. Also, the monthly,seasonal and annual climatologies were detrended. A futuregoal is to apply the detrending for the estimation of the biasesinduced in the temperature data due to instrumental errorssuch as the bathythermographs.

4 Climatological gridded fields

The content of the atlas is described in Sect. 1. For each grid-ded error field, maps are provided that allow one to assess thereliability of the gridded fields and to objectively identify ar-eas with poor coverage. DIVA provides two different errors,the relative (from 0 to 1) and the absolute errors (expressed inphysical parameter units), which depend on the accuracy ofthe observations and their distribution. The relative error hasbeen chosen in this atlas and two threshold values are used(equal to 0.3 or 0.5) for the quality assessment of the results(Fig. 4).

4.1 Description of the gridded fields

While the physical interpretation of the maps is not in thescope of this paper, it is worth noting that the main physicalprocesses are well resolved both in space and time, such asthe following:

1. the clear signature of the temperature and salinity gra-dient from west to east, shown in all monthly, seasonaland annual plots at all depths;

2. the temperature and its seasonal cycle;

3. the Rhodes gyre with the low temperatures is evident inall plots of temperature at surface layers regardless ofmonth or season;

4. the Black Sea outflow with the low temperature andsalinity values, which is evident in the surface plots, re-gardless of month or season.

Figure 5 illustrates the surface temperature at 5 m in Jan-uary, April, July, and October and represents the seasonal cy-cle of the thermal field with a mean shift of 9 ◦C between Jan-uary and July. The western Mediterranean exhibits smallertemperatures than the eastern basin. The differences betweenthe northern and southern regions (meridional gradient) arealso very well represented. The Rhodes Gyre, identified byminimum temperature values, is distinguished southeast ofCrete island, mainly at spring and autumn periods.

Some representative seasonal distributions of salinity at5 m are shown in Fig. 6. The signal of the fresh Atlantic water(S ≈ 36.5 ppt) and its flow along the Algerian coast towardsthe eastern basin through the Sicily Strait is very evident. Wecan also notice the areas of low salinity due to river runoffssuch as the north Adriatic and the north Aegean Sea that isinfluenced by the Dardanelles outflow.

Winter salinity fields at 100 and 500 m respectively areshown in Figs. 3.1 and 3.3 for the last four periods (1970–1979, 1980–1989, 1990–1999, and 2000–2015). There is ahomogeneous increasing trend from west to east comparedto the surface. The salinity range is between about 37 and38.6 ppt at 100 m at the western Mediterranean, while the pe-riods 1980–1989 and 2000–2015 are characterized by higher

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/

A. Iona et al.: Mediterranean Sea Hydrographic Atlas 1295

Figure 11. Comparison of atlas temperature (a, c) and WOA13 climatology (b, d) between January at 0 m (a, b) and September at 1000 m (c,d).

salinities compared to the other periods. Maximum valuesare found in the south Adriatic at 38.9 ppt in the period1980–1989 and 38.8 ppt in the period 2000–2015. In the east-ern Mediterranean, the salinity is between about 38.4 and39.4 ppt at 100 m with the maxima found in the Levantinebasin and Aegean Sea. The last period 2000–2015 is on aver-age more saline than the previous ones, and the thermohalinecirculation patterns (such as the Shikmona Gyre) are moreintense.

In the western basin there is a characteristic increase insalinity at 500 m compared to at 100 m, where it is betweenabout 38.4 and 38.7 ppt. This increase is due to the mix-ing with LIW that flows towards the Strait of Gibraltar. TheTyrrhenian Sea has higher salinities in the periods 1970–1979 and 2000–2015. In the eastern Mediterranean the salin-ity is between about 38.4 and 39.2 ppt, with the highest val-ues in the north and south Aegean Sea and the period 2000–2015 showing higher salinity than the previous years.

4.2 Comparison with SeaDataNet climatology

For the current atlas, a direct comparison with Sea-DataNet V1.1 climatology (Simoncelli et al., 2015a,2016, https://doi.org/10.12770/90ae7a06-8b08-4afe-83dd-ca92bc99f5c0) is performed using statistical indexes likebias (differences between the two climatologies) and RMSE

(root-mean-square error) implemented by Simoncelli et al.(2015a, 2016), during the SeaDataNet 2 project, as con-sistency analysis among different climatological products.The SeaDataNet V1.1 climatology is defined between9.25◦W–36.5◦ E of longitude and 30–46◦ N of latitudewith a horizontal resolution of 1/8◦× 1/8◦ on 30 IODEvertical standard levels from 0 to 4000 m. DIVA softwareversion 4.6.9 (Watelet et al., 2015b) has been used forboth the analysis and background field computation. Thesalinity background field has been computed through annualseminormed analysis considering all available observations,while for temperature, 3-month seminormed backgroundfields centered on the analysis month have been considereddue to the large temperature seasonal variability. Figure 9shows temperature distributions for atlas and SeaDataNetclimatology, for January at 10 m and November at 1000 m.There is very good consistency between both products. Thesame good agreement exists among other months and depths(not shown here). In surface distributions (at 10 m), thedata weighting applied in the atlas (Fig. 9a) has reduced theextension of the influence of Po river.

Statistical indexes like bias (differences between the twoclimatologies) and RMSE (root-mean-square error) showvery good agreement too. Figure 10 shows the basin verti-cal averages of bias and RMSE between the atlas and Sea-DataNet monthly temperature (a) and monthly salinity (b)

www.earth-syst-sci-data.net/10/1281/2018/ Earth Syst. Sci. Data, 10, 1281–1300, 2018

1296 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

climatologies. Higher differences occur in spring and au-tumn, and the atlas is 0.03 ◦C warmer than SeaDataNet inJune and less warm by 0.02 ◦C in November. TemperatureRMSE values are ranging between 0.09 in February and 0.19in July. Maximum differences between the two climatologiesare located in the first layers (not shown here) for all months.The atlas has slightly higher salinity values than SeaDataNetsalinity values in August and December and the mean biasvalue throughout the year is 0.003. Salinity RMSE values areranging between 0.048 and 0.076. As for temperature, maxi-mum differences are located in the first layers for all months.

4.3 Comparison with World Ocean Atlas (WOA13)

WOA13 provides objectively analyzed annual, seasonal, andmonthly fields covering the time period 1955–2012 (aver-age of 6 decadal means) and 6 decades: 1955–1964, 1965–1974, 1975–1984, 1985–1994, 1995–2004, 2005–2012.Each decadal climatology consists of annual (computedas 12-month averages); seasonal (winter: January–March;spring: April–June; summer: July–September; fall: October–December), computed as 3-month averages; and monthlyfields (above 1500 m). Annual, seasonal, and monthly tem-perature and salinity fields are available on 1◦× 1◦ and1/4◦× 1/4◦ latitude–longitude grids. All annual and sea-sonal fields were calculated from 0 to 5500 m depth on102 standard levels. Monthly fields are available only above1500 m on all grids on 57 standard levels. Figure 11 showstemperature fields for the atlas and WOA13 for January at0 m and for September at 1000 m. At the surface, the atlas isable to represent the January mean state with more detail.

The statistical indexes applied to SeaDataNet V1.1 clima-tology have been calculated also for WOA2013, as in Simon-celli et al. (2015a, 2016), and they show a very good agree-ment between both monthly T/S climatologies. Figure 12shows the basin vertical averages of bias and RMSE betweenthe atlas and WOA13 monthly temperature (a) and monthlysalinity (b) climatologies. Temperature bias is always pos-itive, indicating that the atlas is slightly warmer than theWOA13 climatology. Atlas salinity is less than the WOA13at an average of 0.007 throughout the whole year.

4.4 Comparison with Medar/Medatlas Climatology

The Medar/Medatlas climatology was computed using thevariational inverse method as the current atlas. Correlationlength was fixed a priori to a constant value according tothe a priori knowledge of typical scales of parameters in thedomain of interest. The signal-to-noise ratio was calibratedby a generalized cross-validation technique. Data closer than15 km from the coast or in areas shallower than 50 m wererejected in order to avoid the influence of coastal features onopen sea (Rixen et al., 2005).

The impact of such as rejection can be seen in Fig. 13:in the north Adriatic Sea and in the north Aegean Sea, the

Figure 12. Basin vertical averages of BIAS and RMSE betweenatlas and WOA13 monthly temperature (a), and monthly salinity (b)climatologies.

influence of Po river and Black Sea outflows during winter isnot captured.

The temperature at 10 m for the decade 1991–2000 isshown below for the atlas (Fig. 14a) and the Medar/Medatlasclimatology (Fig. 14b). The Medar/Medatlas profiles extenduntil 2000, so this decade was chosen as the most completeone since there is no error field available to mask regionsempty of data.

As it can be seen in Fig. 14, the current atlas describes withmore detail the several features of the general circulation pat-tern.

5 Code and data availability

The dataset used in this work (Simoncelliet al., 2015b, http://sextant.ifremer.fr/record/8c3bd19b-9687-429c-a232-48b10478581c/) is avail-able from the SeaDataNet catalogue at https://www.seadatanet.org/Products. Updated versions willbe released periodically.

The DIVA source code is distributed via GitHub athttps://github.com/gher-ulg/DIVA (last access: 9 July 2018).

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/

A. Iona et al.: Mediterranean Sea Hydrographic Atlas 1297

Figure 13. January temperature at 10 m: (a) Medar/Medatlas, coastal data removed from the analysis; (b) data used (imported from http://modb.oce.ulg.ac.be/backup/medar/JPGSTATIONS/medar.01.med.temp.23.3.0.stations.jpg, last access: 9 July 2018); (c) current climatology,coastal data included; (d) data used.

Figure 14. (a) Atlas masked temperature field for the decade 1991–2000 at 10 m, (b) Medar/Medatlas temperature field for the decade1991–2000, at 10 m (no error field is available).

The different versions of the software tool are archivedand referenced in Zenodo platform under the DOI:https://doi.org/10.5281/zenodo.592476.

The atlas itself is distributed through Zenodo ac-cording to the following subproducts: annual cli-matology: https://doi.org/10.5281/zenodo. 1146976(Iona, 2018a); seasonal climatology for 57 run-ning decades from 1950-1959 to 2006–2015:https://doi.org/10.5281/zenodo.1146938 (Iona, 2018b); sea-sonal climatology: https://doi.org/10.5281/zenodo.1146953

(Iona, 2018c). annual climatology for 57 run-ning decades from 1950–1959 to 2006–2015:https://doi.org/10.5281/zenodo.1146957 (Iona, 2018d);seasonal climatology for six periods: 1950–1959,1960–1969, 1970–1979, 1980–1989, 1990–1999, 2000–2015: https://doi.org/10.5281/zenodo.1146966 (Iona,2018e); annual climatology for six periods: 1950–1959, 1960–1969, 1970–1979, 1980–1989, 1990–1999,2000–2015: https://doi.org/10.5281/zenodo.1146970

www.earth-syst-sci-data.net/10/1281/2018/ Earth Syst. Sci. Data, 10, 1281–1300, 2018

1298 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

(Iona, 2018f); monthly climatology:https://doi.org/10.5281/zenodo.1146974 (Iona, 2018g).

6 Conclusions

A new, high-resolution atlas of temperature and salinity forthe Mediterranean Sea for the period 1950–2015 is pre-sented. The analysis is based on the latest SeaDataNet V2dataset (Simoncelli et al., 2015b, http://sextant.ifremer.fr/record/8c3bd19b-9687-429c-a232-48b10478581c/), provid-ing the most complete, extended, and improved collection ofin situ observations.

The results describe well the expected distributions ofthe hydrological characteristics and reveal their long-termchanges. Techniques for overcoming the inhomogeneousvertical and spatial distribution of oceanographic mea-surements were presented. The analysis focused on thedata of 1950 and onwards where more reliable instrumen-tal observations exist. Comparisons with SeaDataNet V1.1climatology (Simoncelli et al., 2015a, 2016, http://doi.org/10.12770/90ae7a06-8b08-4afe-83dd-ca92bc99f5c0) re-veal a very good agreement. It was expected since both twoclimatologies use similar interpolation methodology. The at-las offers additional value-added decadal temperature andsalinity distributions which already exist in the region fromprevious versions of Medar/Medatlas climatology, thoughwith no error fields available. This is of particular impor-tance.

There are still improvements to be implemented in futureversions of the atlas, such as a three-dimensional analysisinstead of stacking the two-dimensional horizontal levels to-gether in order to take into account the vertical correlationof the parameters. This would reduce vertical inconsisten-cies that may remain in the results. In addition, the correc-tion of the warm bias in the bathythermograph data, causedby instrumental errors, should also be addressed. However,it is anticipated that the approach followed here concerningthe calibration of the analysis parameters will be followedby other groups in the future for the Mediterranean climatestudies and other applications related with the long-term vari-ability of the hydrological characteristics of the region and itsclimate change.

Another future improvement is the use of a dynamical con-straint such as a real velocity field or an advection constraintbased on topography that would allow anisotropies in the cor-relations to be included into the analysis. This atlas aims tocontribute to existing available knowledge in the region andfill existing data gaps in space and time.

Author contributions. AI created the atlas, wrote the first versionof the paper and prepared the figures. SS provided the MATLABprograms to compute the statistical indexes and plot them. JMB,SW, CT and AT reviewed the paper. CT and SW formatted the doc-ument in LaTeX.

Competing interests. The authors declare that they have no con-flict of interest.

Disclaimer. It cannot be warranted that the atlas is free from errorsor omissions. Correct and appropriate atlas interpretation and usageis solely the responsibility of data users.

Acknowledgements. The DIVA development has receivedfunding from the European Union Sixth Framework Programme(FP6/2002–2006) under grant agreement no. 026212, Sea-DataNet (http://www.seadatanet.org/, last access: 9 July 2018),the Seventh Framework Programme (FP7/2007–2013) undergrant agreement no. 283607, SeaDataNet II, SeaDataCloud andEMODnet (http://www.emodnet.eu/, last access: 9 July 2018)(MARE/2008/03 – Lot 3 Chemistry – SI2.531432) from thehttp://ec.europa.eu/dgs/maritimeaffairs_fisheries/index_en.htm(last access: 9 July 2018) Directorate-General for Maritime Affairsand Fisheries.

Edited by: Giuseppe M. R. ManzellaReviewed by: Alexey Mishonov and one anonymous referee

References

Beckers, J.-M., Barth, A., Troupin, C., and Alvera-Azcárate,A.: Some approximate and efficient methods to assess er-ror fields in spatial gridding with DIVA (Data InterpolatingVariational Analysis), J. Atmos. Ocean. Tech., 31, 515–530,https://doi.org/10.1175/JTECH-D-13-00130.1, 2014.

Bergamasco, A. and Malanotte-Rizzoli, P.: The circulation of theMediterranean Sea: a historical review of experimental investi-gations, Advances in Oceanography and Limnology, 1, 11–28,https://doi.org/10.1080/19475721.2010.491656, 2010.

Béthoux, J.-P., Gentili, B., and Tailliez, D.: Warming and freshwaterbudget change in the Mediterranean since the 1940s, their pos-sible relation to the greenhouse effect, Geophys. Res. Lett., 25,1023–1026, https://doi.org/10.1029/98gl00724, 1998.

Brankart, J.-M. and Brasseur, P.: Optimal analysis ofIn Situ Data in the Western Mediterranean UsingStatistics and Cross-Validation, J. Atmos. Ocean.Tech., 13, 477–491, https://doi.org/10.1175/1520-0426(1996)013<0477:OAOISD>2.0.CO;2, 1996.

Brankart, J.-M. and Brasseur, P.: The general circulation inthe Mediterranean Sea: a climatological approach, J. MarineSyst., 18, 41–70, https://doi.org/10.1016/S0924-7963(98)00005-0, 1998.

Brasseur, P., Beckers, J., Brankart, J., and Schoenauen, R.: Seasonaltemperature and salinity fields in the Mediterranean Sea: Clima-tological analyses of a historical data set, Deep-Sea Res. Pt. I, 43,159–192, https://doi.org/10.1016/0967-0637(96)00012-x, 1996.

Bretherton, F., Davis, R., and Fandry, C.: A technique for objectiveanalysis and design of oceanic experiments applied to Mode-73, Deep-Sea Res., 23, 559–582, https://doi.org/10.1016/0011-7471(76)90001-2, 1976.

Capet, A., Troupin, C., Carstensen, J., Grégoire, M., and Beckers,J.-M.: Untangling spatial and temporal trends in the variability of

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/

A. Iona et al.: Mediterranean Sea Hydrographic Atlas 1299

the Black Sea Cold Intermediate Layer and mixed Layer Depthusing the DIVA detrending procedure, Ocean Dynam., 64, 315–324, https://doi.org/10.1007/s10236-013-0683-4, 2014.

Gandin, L. S.: Objective analysis of meteorological fields. By L. S.Gandin. Translated from the Russian, Jerusalem (Israel Programfor Scientific Translations), 1965, Q. J. Roy. Meteor. Soc., 92,447–447, https://doi.org/10.1002/qj.49709239320, 1966.

Guibout, P.: Atlas Hydrologique de la Mediterranee, Shom, Ifremer,Paris, France, 1 Edn., 150 pp., 1987.

Hecht, A., Pinardi, N., and Robinson, A. R.: Currents, WaterMasses, Eddies and Jets in the Mediterranean Levantine Basin, J.Phys. Oceanogr., 18, 1320–1353, https://doi.org/10.1175/1520-0485(1988)018<1320:CWMEAJ>2.0.CO;2, 1988.

Iona, A.: Mediterannean Sea-Temperature and Salin-ity Annual Climatology [Data set], Zenodo,https://doi.org/10.5281/zenodo.1146976, 2018a.

Iona, A.: Mediterannean Sea-Temperature and Salin-ity Seasonal Climatology for 57 running decadesfrom 1950–1959 to 2006–2015 [Data set], Zenodo,https://doi.org/10.5281/zenodo.1146938, 2018b.

Iona, A.: Mediterannean Sea-Temperature and Salin-ity Seasonal Climatology [Data set], Zenodo,https://doi.org/10.5281/zenodo.1146953, 2018c.

Iona, A.: Mediterannean Sea-Temperature and Salinity Annual Cli-matology for 57 running decades from 1950–1959 to 2006–2015[Data set], Zenodo, https://doi.org/10.5281/zenodo.1146957,2018d.

Iona, A.: Mediterannean Sea-Temperature and Salinity SeasonalClimatology for six periods (1950–1959, 1960–1969, 1970–1979, 1980–1989, 1990–1999, 2000–2015) [Data set], Zenodo,https://doi.org/10.5281/zenodo.1146966, 2018e.

Iona, A.: Mediterannean Sea-Temperature and Salinity AnnualClimatology for six periods (1950–1959, 1960–1969, 1970–1979, 1980–1989, 1990–1999, 2000–2015) [Data set], Zenodo,https://doi.org/10.5281/zenodo.1146970, 2018f.

Iona, A.: Mediterannean Sea-Temperature and Salin-ity Monthly Climatology [Data set], Zenodo,https://doi.org/10.5281/zenodo.1146974, 2018g.

Klein, B., Roether, W., Manca, B. B., Bregant, D., Beitzel, V., Ko-vacevic, V., and Luchetta, A.: The large deep water transient inthe Eastern Mediterranean, Deep-Sea Res. Pt. I, 46, 371–414,https://doi.org/10.1016/s0967-0637(98)00075-2, 1999.

Lascaratos, A., Roether, W., Nittis, K., and Klein, B.: Recentchanges in deep water formation and spreading in the east-ern Mediterranean Sea: a review, Prog. Oceanogr., 44, 5–36,https://doi.org/10.1016/s0079-6611(99)00019-1, 1999.

La Violette, P. E.: The Western Mediterranean Circulation Ex-periment (WMCE): Introduction, J. Geophys. Res., 95, 1511,https://doi.org/10.1029/jc095ic02p01511, 1990.

Levitus, S.: Climatological Atlas of the World Ocean, Tech. rep.,NOAA Professional Paper 13, US. Government Printing Office:Washington, DC, 1982.

Locarnini, R. A., Mishonov, A. V., Antonov, J. I., Boyer, T. P.,Garcia, H. E., Baranova, O. K., Zweng, M. M., Paver, C. R.,Reagan, J. R., Johnson, D. R., Hamilton, M., and Seido, D.:World Ocean Atlas 2013, Volume 1: Temperature, in: NOAAAtlas NESDIS, edited by: Levitus, S. and Mishonov A., 73,40 pp., NOAA, available at: http://data.nodc.noaa.gov/woa/WOA13/DOC/woa13_vol1.pdf (last access: 9 July 2018), 2013.

Lozier, M. S., Owens, W. B., and Curry, R. G.: The clima-tology of the North Atlantic, Prog. Oceanogr., 36, 1–44,https://doi.org/10.1016/0079-6611(95)00013-5, 1995.

Malanotte-Rizzoli, P. and Hecht, A.: Large-scale properties of theeastern mediterranean: a rewiew, Oceanol. Acta, 11, 323–335,1988.

Malanotte-Rizzoli, P., Manca, B. B., d’Alcala, M. R., Theocharis,A., Brenner, S., Budillon, G., and Ozsoy, E.: The EasternMediterranean in the 80s and in the 90s: the big transition inthe intermediate and deep circulations, Dynam. Atmos. Oceans,29, 365–395, https://doi.org/10.1016/s0377-0265(99)00011-1,1999.

Manca, B., Burca, M., Giorgetti, A., Coatanoan, C., Garcia, M.-J.,and Iona, A.: Physical and biochemical averaged vertical pro-files in the Mediterranean regions: an important tool to tracethe climatology of water masses and to validate incoming datafrom operational oceanography, J. Marine Syst., 48, 83–116,https://doi.org/10.1016/j.jmarsys.2003.11.025, 2004.

Manzella, G. M. R. and Gambetta, M.: Implementation of real-timequality control procedures by means of a probabilistic estimateof seawater temperature and its temporal evolution, J. Atmos.Ocean. Tech., 30, 609–625, https://doi.org/10.1175/JTECH-D-11-00218.1, 2013.

MEDAR Group: MEDATLAS/2002 database, Mediterranean andBlack Sea database of temperature salinity and bio-chemical pa-rameters, Climatological Atlas, Tech. rep., Ifremer, 4 Cdroms,2002.

Millot, C.: Circulation in the Western Mediterranean Sea, Oceanol.Acta, 10, 143–148, 1987.

Millot, C.: Mesoscale and seasonal variabilities of the circulationin the western Mediterranean, Dynam. Atmos. Oceans, 15, 179–214, https://doi.org/10.1016/0377-0265(91)90020-G, 1991.

Millot, C.: Circulation in the Western Mediterranean Sea,J. Marine Syst., 20, 423–442, https://doi.org/10.1016/S0924-7963(98)00078-5, 1999.

Ovchinnikov, I.: Circulation in the surface and intermediate layersof the mediterranean, Oceanology, 6, 49–59, 1966.

Picco, P.: Climatological Atlas of the Western Mediterranean, Tech.rep., ENEA Santa Teresa Centre, La Spezia, Italy, 224 pp., 1990.

Rahmstorf, S.: Influence of mediterranean outflow onclimate, EOS T. Am. Geophys. Un., 79, 281–281,https://doi.org/10.1029/98eo00208, 1998.

Reiniger, R. and Ross, C.: A method of interpolation with ap-plication to oceanographic data, Deep-Sea Res., 15, 185–193,https://doi.org/10.1016/0011-7471(68)90040-5, 1968.

Rixen, M., Beckers, J.-M., Brankart, J.-M., and Brasseur, P.: A nu-merically efficient data analysis method with error map gener-ation, Ocean Modell., 2, 45–60, https://doi.org/10.1016/S1463-5003(00)00009-3, 2000.

Rixen, M., Beckers, J.-M., and Allen, J.: Diagnosis of vertical veloc-ities with the QG Omega equation: a relocation method to obtainpseudo-synoptic data sets, Deep-Sea Res. Pt. I, 48, 1347–1373,https://doi.org/10.1016/S0967-0637(00)00085-6, 2001.

Rixen, M., Beckers, J.-M., Levitus, S., Antonov, J., Boyer, T.,Maillard, C., Fichaut, M., Balopoulos, E., Iona, S., Dooley,H., Garcia, M.-J., Manca, B., Giorgetti, A., Manzella, G.,Mikhailov, N., Pinardi, N., Zavatarelli, M., and the Medar Con-sortium: The Western Mediterranean Deep Water: a proxy

www.earth-syst-sci-data.net/10/1281/2018/ Earth Syst. Sci. Data, 10, 1281–1300, 2018

1300 A. Iona et al.: Mediterranean Sea Hydrographic Atlas

for global climate change, Geophys. Res. Lett., 32, L12608,https://doi.org/10.1029/2005GL022702, 2005.

Robinson, A., Leslie, W., Theocharis, A., and Lascaratos, A.:Mediterranean Sea Circulation, Encyclopedia of Ocean Sciences,1689–1705, https://doi.org/10.1006/rwos.2001.0376, 2001.

Roether, W., Manca, B. B., Klein, B., Bregant, D., Georgopoulos,D., Beitzel, V., Kovacevic, V., and Luchetta, A.: Recent Changesin Eastern Mediterranean Deep Waters, Science, 271, 333–335,https://doi.org/10.1126/science.271.5247.333, 1996.

Schlitzer, R.: Interactive analysis and visualization of geosciencedata with Ocean Data View, Comput. Geosci., 28, 1211–1218,https://doi.org/10.1016/S0098-3004(02)00040-7, 2002.

Schröder, K., Gasparini, G. P., Tangherlini, M., and Astraldi, M.:Deep and intermediate water in the western Mediterranean underthe influence of the Eastern Mediterranean Transient, Geophys.Res. Lett., 33, https://doi.org/10.1029/2006gl027121, 2006.

Schroeder, K., Chiggiato, J., Bryden, H. L., Borghini, M.,and Ben Ismail, S.: Abrupt climate shift in the West-ern Mediterranean Sea, Scientific Reports, 6, 23009,https://doi.org/10.1038/srep23009, 2016.

SeaDataNet Group: Data Quality Control Procedures, Tech.Rep. Version 2.0, SeaDataNet consortium, availabel at:https://www.seadatanet.org/content/download/596/3118/file/SeaDataNet_QC_procedures_V2_(May_2010).pdf?version=1,(last access: 9 July 2018), 2010.

Simoncelli, S., Tonani, M., Grandi, A., Coatanoan, C., Myroshny-chenko, V., Sagen, H., Back, O., Scory, S., Schlitzer, R., andFichaut, M.: First Release of the SeaDataNet Aggregated DataSets Products, WP10 Second Year Report – DELIVERABLED10.2, https://doi.org/10.13155/49827, 2014.

Simoncelli, S., Coatanoan, C., Myroshnychenko, V., Sagen, H.,Back, O., Scory, S., Grandi, A., Barth, A., and Fichaut, M.:SeaDataNet, First Release of Regional Climatologies, WP10Third Year Report – DELIVERABLE D10.3, available at:https://doi.org/10.13155/50381, 2015a.

Simoncelli, S., Coatanoan, C., Myroshnychenko, V., Sagen, H.,Back, O., Scory, S., Grandi, A., Schlitzer, R., and Fichaut,M.: Second release of the SeaDataNet aggregated data setsproducts, WP10 Fourth Year Report – DELIVERABLE D10.4,https://doi.org/10.13155/50382, 2015b.

Simoncelli, S., Grandi, A., and Iona, S.: New MediterraneanSea climatologies, Bollettino di Geofisica, Vol. 57 – Supple-ment, 2016, 252 pp., available at: https://imdis.seadatanet.org/Previous-editions/IMDIS-2016/Proceedings (last access: 9 July2018), 2016.

Troupin, C., Machín, F., Ouberdous, M., Sirjacobs, D., Barth,A., and Beckers, J.-M.: High-resolution Climatology ofthe North-East Atlantic using Data-Interpolating Varia-tional Analysis (Diva), J. Geophys. Res., 115, C08005,https://doi.org/10.1029/2009JC005512, 2010.

Troupin, C., Sirjacobs, D., Rixen, M., Brasseur, P., Brankart, J.-M., Barth, A., Alvera-Azcárate, A., Capet, A., Ouberdous, M.,Lenartz, F., Toussaint, M.-E., and Beckers, J.-M.: Generationof analysis and consistent error fields using the Data Interpolat-ing Variational Analysis (Diva), Ocean Modell., 52-53, 90–101,https://doi.org/10.1016/j.ocemod.2012.05.002, 2012.

Troupin, C., Watelet, S., Barth, A., and Beckers, J.-M.: Diva-User-Guide: v1.0 (Version v1.0). Zenodo., Tech. rep., GeoHydrody-namics and Environment Research, University of Liège, Liège,Belgium, https://doi.org/10.5281/zenodo.836723, 2017.

Watelet, S., Ouberdous, M., Troupin, C., Barth, A., and Beckers, J.-M.: DIVA Version 4.6.11, Tech. rep., GeoHydrodynamics andEnvironment Research, University of Liège, Liège, Belgium,https://doi.org/10.5281/zenodo.400970, 2015a.

Watelet, S., Ouberdous, M., Troupin, C., Barth, A., and Beckers,J.-M.: DIVA Version 4.6.9, Tech. rep., GeoHydrodynamics andEnvironment Research, University of Liège, Liège, Belgium,https://doi.org/10.5281/zenodo.400968, 2015b.

Watelet, S., Troupin, C., Beckers, J.-M., Barth, A., and Ouber-dous, M.: DIVA Version 4.7.1, Tech. rep., GeoHydrodynamicsand Environment Research, University of Liège, Liège, Belgium,https://doi.org/10.5281/zenodo.836727, 2017.

Wijffels, S. E., Willis, J., Domingues, C. M., Barker, P., White,N. J., Gronell, A., Ridgway, K., and Church, J. A.: ChangingExpendable Bathythermograph Fall Rates and Their Impact onEstimates of Thermosteric Sea Level Rise, J. Climate, 21, 5657–5672, https://doi.org/10.1175/2008jcli2290.1, 2008.

Zweng, M., Reagan, J., Antonov, J., Locarnini, R., Mishonov, A.,Boyer, T., Garcia, H., Baranova, O., Johnson, D., Seidov, D.,and Biddle, M.: World Ocean Atlas 2013, Volume 2: Salinity, in:NOAA Atlas NESDIS, edited by: Levitus, S. and Mishonov, A.,74, 39 pp., NOAA, available at: http://data.nodc.noaa.gov/woa/WOA13/DOC/woa13_vol2.pdf (last access: 9 July 2018), 2013.

Earth Syst. Sci. Data, 10, 1281–1300, 2018 www.earth-syst-sci-data.net/10/1281/2018/